views:

30

answers:

2

We want to monitor some processes on windows machine. If process is down for 30 minutes, an alarm will be raised. Is it possible/or necessary to monitor a process downtime precisely, say, a process is down EXACTLY for 30 minutes, then an alarm will be raised? Normally, we can check it every 1 minute, but technically, most of the time, you could miss somes seconds.

A: 

Look into Nagios and NSClient++ if you can monitor your Windows host from a Linux host. If this would be an option there is room to do this and many many other things in regards to monitoring your processes and more.

gnucom
+2  A: 

If you're looking for something much lighter and simpler then Nagios, see AlertGrid. It is very esay to use, the only downside is that it requires a bit of integration: AlertGrid only LISTENS for heartbeat signals, so you have to provide them manually (the API is extremely simple).

The other cool thing is that if the process you want to monitor runs YOUR code - you can send heartbeat events directly "from inside", and these events can carry your own custom parameters. Then, in the AlertGrid - you can easily manage custom rules around these parameters. So if the executable you monitor is, for instance, an order processing application you can send parameter called 'number_of_orders_processed' and create rule "if number_of_orders_processed > 100, send SMS message / make phone call to... " and it will work immediately.

I am in the AlertGrid dev team, if you have any questions - feel free to ask.

PawelRoman